You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, TiFlash will create a RNFetchPagesWorker and PrepareStreamsWorker for each query. This introduces the cost of frequently creating and destroying threads, and at the same time, it is impossible to control the number of thread concurrency on a node, which can easily lead to OOM.
We can use RNFetchPagesWorker and PrepareStreamsWorker as global thread pools and shard these thread pools between queries.
The text was updated successfully, but these errors were encountered:
Enhancement
Currently, TiFlash will create a
RNFetchPagesWorker
andPrepareStreamsWorker
for each query. This introduces the cost of frequently creating and destroying threads, and at the same time, it is impossible to control the number of thread concurrency on a node, which can easily lead to OOM.We can use
RNFetchPagesWorker
andPrepareStreamsWorker
as global thread pools and shard these thread pools between queries.The text was updated successfully, but these errors were encountered: